To evaluate the concordance between register-derived exposure estimates and biological detection of medications using updated methods.
Approach:
Population and Data Sources: Data from the Swedish National Board of Forensic Medicine and nationwide health registers were linked to assess forensically investigated deaths.
Methodology: The PRE2DUP method was applied to reconstruct continuous drug exposure periods from dispensing data, focusing on the time closest to death.
Analysis: Agreement between PRE2DUP-modelled drug exposure and toxicological results was evaluated using sensitivity, specificity, predicted adherence, and Cohen's κ.
Key Findings:
The updated analysis included 10,060 individuals for substances with consistent testing and 18,627 for those tested throughout the study period.
Predicted adherence and Cohen's κ showed significant improvements for several drug classes, particularly cardiovascular drugs.
Antidepressants and antiepileptics also demonstrated moderate increases in agreement metrics.
Limitations:
Observations from periods with inconsistent testing were excluded, potentially limiting the generalizability of findings.
The study focused on specific substances, which may not represent all medications.
